Influence of social networks on cancer survivors' self‐management support: A mixed methods study

Abstract Objective The role of social networks, especially weaker ties (e.g. casual acquaintances and hobby groups), in self‐management of long‐term consequences of cancer is unexplored. This study aimed to explore the structure of cancer survivors' social networks and their contribution to self‐management support and health‐related quality of life (HRQoL). Methods The study used a sequential, exploratory mixed methods design. Phase 1 surveyed 349 lymphoma, colorectal, breast and prostate cancer survivors. Phase 2 analysed 20 semi‐structured interviews with respondents recruited from Phase 1. Results Phase 1 results suggested participants' HRQoL increased if they participated in an exercise group, if their self‐management skills increased, and social distress and negative illness perception decreased (p < 0.0005 adj. R 2  = 0.631). These findings were explored in Phase 2, identifying underlying mechanisms. Four themes were identified: disrupted networks after cancer treatment; navigating formal support and building individual capacity; peer networks and self‐management knowledge and linking networks to enable adaptation in recovery. Conclusions This study suggests engagement with community groups, particularly those not directly related to illness management and social interaction with weak ties, make a valuable contribution to self‐management support, increase HRQoL and enhance well‐being.


| INTRODUCTION
The role that social networks and connections play in shaping a person's behaviour and subsequent impact on health and well-being is increasingly recognised. Seminal studies demonstrated relationships between increased social engagement and reductions in mortality and morbidity (Berkman & Syme, 1979;Christakis & Fowler, 2007;House et al., 1988). The influence of social networks on self-management support for long-term conditions, such as diabetes, suggests that having a diverse network increases social connectedness and satisfaction with current networks and associated with enhanced self-management skills, physical and mental well-being (Vassilev et al., 2016). This may in part be due to diverse social networks providing greater access to informal practical resources (Kroenke et al., 2013). However, larger social networks and networks including different types of relationships can also require higher levels of relationship management (Vassilev et al., 2019) and be emotionally burdensome to manage. This indicates that the underlying mechanisms through which social networks operate are complex and that networks may have negative, as well as positive, impacts on health and quality of life (Cheng et al., 2013;Hamilton et al., 2010;Vassilev et al., 2016).
Previous cancer social network studies, largely undertaken in women with breast cancer, have found associations between increased network size, overall survival and cancer survival (Beasley et al., 2010;Jones & Storksdieck, 2019;Kroenke et al., 2006Kroenke et al., , 2017Lindstrom & Rosvall, 2019;Sarma et al., 2018;Waxler-Morrison et al., 1991). Other studies have identified relationships between higher levels of social network engagement and higher HRQoL (Cheng et al., 2013;Kroenke et al., 2013;Lim & Zebrack, 2006;Soares et al., 2013), lower inflammatory markers and depressive symptoms (Hughes et al., 2014), increased exercise engagement (Kim et al., 2015) and increased support for healthy eating (Crookes et al., 2016). Social support has been found to be valuable in self-management, but there is limited research exploring network characteristics or utilising social network approaches and theories (Balfe et al., 2017;Henshall et al., 2018;Kim et al., 2020;Paterson et al., 2015). Few studies have examined how Socio-Economic Status (SES) could influence social network access to self-management resources for people with cancer or long-term conditions (Juárez-Ramírez et al., 2015;. As the incidence of those living with and beyond cancer is predicted to rise (Maddams et al., 2012), self-management support has been adopted as an approach to meet increasing health and wellbeing needs of cancer survivors (Batehup et al., 2017). Selfmanagement has limitations as it frequently focuses on individual concerns, such as relapse (Fenlon et al., 2015) and does not consider how personal agency, shaped by social networks can influence selfmanagement behaviour outcomes (Dunn et al., 2021).
Drawing on social network theories (Figure 1), the aims of this study were to contribute towards the development of a contextualised understanding of self-management and self-management support by

| Phase 1: Survey
Over 11 months 621 participants meeting the eligibility criteria were approached across 5 Trusts, with 349 consenting to Phase 1 of the study, eliciting a 56% response rate (Burns & Grove, 2003) (Figure 2).
The survey consisted of five components:

| Health related quality of life
HRQoL data were collected using the Functional Assessment of Cancer Therapy-General (FACT-G) (Cella et al., 1993) because it is generalisable to all people with a cancer diagnosis.

| Participant social engagement and characteristics
Demographic data were collected on cancer diagnosis and treatment.
SES was determined using the Indices of Multiple Deprivation, derived from participants' postcode (Gov.uk, 2015). Engagement in social activity data, for example, recreation, were collected using a section from a questionnaire developed by Vassilev et al. (2013) and previously used with participants with long-term conditions. Face validity was gained through Patient and Public Involvement (PPI) review and pilot study (Table 1).

| Social network characteristics
A social network assessment tool was developed based on the name generator approach (Vassilev et al., 2013). Participants identified all members of their social network by name, describing their relationship to them (e.g. 'Sarah' and 'work friend'). Participants F I G U R E 2 Flow chart of recruitment of participants in the study scored each member's contribution to their self-management support (0 = never to 5 = a lot) in three domains: illness work (e.g. managing medication), day to day work (e.g. housework) and emotional work (e.g. someone to talk to about worries) and indicated how close members lived to them, for example, a short walk. Face validity was achieved through PPI review and pilot study.

| Self-management
Data were collected using subsection 4 'self-monitoring and insight' and 6 'skill and technique acquisition' from the Health Education Impact Questionnaire (HEIQ) (Osborne et al., 2007), previously validated in cancer populations (Maunsell et al., 2014).

| Social distress
Data were collected using the Brief Illness Perception Questionnaire (BPIQ) (Broadbent et al., 2006), adapted with permission for use in a cancer population. Face validity was gained through feedback from PPI group and pilot study. Cronbach's alpha was undertaken to test internal reliability. The result was 0.68, under the recommended 0.7, but lower results are acceptable within psychological constructs (Klein, 1999). Social Distress data were collected using the Social Difficulties Inventory (SDI) (Wright et al., 2011).

| Statistical analyses
Descriptive statistics are used to summarise demographic and social network dimensions of the sample. Preliminary analyses were  (Kvale, 2007;Quin Patton, 2002;. Techniques to enhance trustworthiness of data collection were addressed by (R. A.) and (I. V.) critically reviewing a sample of interview recordings (Lincoln & Guba, 1985).
The dynamic relationship between the interviewer and interviewees was acknowledged. The interviewer kept a critical journal to reflect on each interview and minimise researcher influence (Bryman, 2016;Spradley, 1979).

| Data analyses and interpretation
Interview findings were analysed deductively, using framework analysis (Gale et al., 2013). Themes identified from survey findings and informed by weak tie theory (Granovetter, 1973) were used to describe and explain how weak tie (non-familial and peripheral) social network members, social network mechanisms and SES influenced self-management support. Data were imported into the software programme NVivo 12 to facilitate analysis (Ritchie & Lewis, 2014).
G. H. J. conducted the analysis; subsamples of which were independently coded by A. R. and I. V.

| Data integration
Integration occurred throughout the study.   Survey findings directed analysis aims of Phase 2 of the study.
Interview data enabled exploration of why membership of social network groups might contribute towards lower social distress, increased self-management and HRQoL and elicit mechanisms through which this occurred. Analysis also gave the opportunity to explore the subtle influence of SES on social networks and self-management.

| Phase 2 findings
A purposeful sample of 20 participants was recruited from the 220 survey participants who expressed an interest in being interviewed ( Close family members are frequently turned to in times of acute need (Perry & Pescosolido, 2015) and our findings concur but also identified for the first time the mechanisms of how weak tie networks, such as recreational groups, informal peer networks and community cancer support services make valuable contributions to selfmanagement and lowering social distress after treatment. The value of network groups, particularly peer support has been recognised (Dunn et al., 2003). Our paper contextualises the contribution of peer support to self-management within the wider focus of a network approach, while offering some insights as to why support groups may not work for all and the importance of other weak tie networks.

| Network membership, socio-economic status and self-management support
Our findings suggest that characteristics of network membership, such as diversity, appeared to have more influence on ability to engage with network self-management opportunities and that lower SES alone was not prohibitive of network group engagement. Phase 2 findings revealed that participants whose networks were dominated by family members tended to have low SES. Phase 2 findings also indicated that while low SES did not appear to restrict access to direct self-management support it did appear to be indirectly associated with preventing access to social resources, such as transport, limiting engagement. Participants with higher SES were less impacted by access to resources and benefited from additional resources, such as occupational health services.
Participants with lower SES and family dominated networks appeared to have limited opportunities to engage with wider selfmanagement support and resources. This could be due to limited individual and network resources (e.g. employment flexibility) and the potential burden that self-management support could put on network members (Kroenke et al., 2013;Perry & Pescosolido, 2015;Walker et al., 2018). It is possible that participants who had embedded family networks could have been satisfied with the variety of resources already provided within their network and did not feel it necessary to seek external support. The substantial emotional and identity investment people have within families can make it challenging to renegotiate these relationships and engage in new or alternative self-management resources (Vassilev et al., 2016). This is more difficult during a time of crisis when people's networks tend to shrink even though self-management and quality of life might benefit from access to larger and more diverse networks.
Acknowledging the contribution of wider social networks and understanding how relationships, positive or negative, within the context of such networks may shape one another, could potentially contribute to upscaling cancer survivorship care, bridging the gap between hospital self-management support and community social networks.

| Methodological value
The study indicated the value of adopting a mixed method network approach to illuminate self-management support of cancer survivors.
Findings suggest the influence of structural components of networks, such as size and diversity have a nuanced influence on how and why cancer survivors access or do not access resources and support, which cannot be captured and explained by only drawing on either quantitative or qualitative methods alone. Our findings also indicate that HRQoL may be too crude an outcome measure to identify the influence of social networks on cancer survivors' social distress and ability to self-manage. Utilising a well-being outcome measure in conjunction with HRQoL would give the opportunity to measure a variety of positive assets in functioning.
Findings also indicate data collection methods and measures used in the quantitative phase of the study tended to underestimate the involvement and role of weak ties in social networks and selfmanagement support, which could reflect the value cancer survivors' put on relationships with stronger ties. The value of adopting a mixed method approach was further demonstrated as the qualitative analysis demonstrated the key role weak ties play in self-management support and quality of life is largely invisible and their value is precisely due to these links being understated by participants .
Developing and using measures of network engagement and support capable of capturing the role of weak ties and relationship work is likely to lead to a better understanding of the needs and experiences of cancer survivors.

| Limitations
We recognise that a more robust sequential approach could have been adopted whereby results of the survey directly informed the interview schedule, and not just the analysis. Parallel data collection limited the strengths gained from adopting a mixed method design and could have impacted study findings. Participants were predominantly white (2% self-identified as BME), educated, middle income and did not reflect greater ethnic diversity or a broader SES population.
The HRQoL measure Fact-G (Cella et al., 1993) has been criticised as not reflecting the broader concerns of cancer survivors who have completed treatment (Yost et al., 2013). Future studies could consider more cancer survivor specific HRQoL measures such as the Quality of life in Adult Cancer Survivors (Avis et al., 2005). These limitations suggest the findings need to be interpreted with caution in terms of implications for the wider population.

| CONCLUSIONS AND CLINICAL IMPLICATIONS
This study set out to explore if the dimensions of cancer survivors' personal social networks shaped self-management support and their relationship with HRQoL. Taking a mixed methods approach and drawing on social network theories, findings suggest that engagement in community groups and interactions with weak tie social network members can make a substantial contribution to self-management and improve HRQoL. This study identified the previously under-

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.